Update transformers version to 4.54.1 to fix Qwen3 rotary embedding dimension mismatch 4f4a3c9 Florian valade commited on 10 days ago
Fix transformers compatibility: pin versions and rename past_key_value to past_key_values 687049b Florian valade commited on 10 days ago
Track metrics during streaming, remove redundant generation re-runs 33efa44 Florian valade commited on 10 days ago
Fix early exit inference loop to eliminate redundant computation a781577 Florian valade commited on 11 days ago
Fix: remove undefined use_local argument from get_decoder call 45e00e6 Florian valade commited on 21 days ago